野生动物监测无线传感器网络中多目标被动式定位关键技术研究

In the protection of wildlife, being lack of long-term scientific data accumulation, people can't accurately be aware of the migration, loss, disease, death, and poaching for the wild animals. Target locating technology from Wireless Sensor Network (WSN) can be used to detect the position of animals. Because most of the area where wildlife live are human inaccessible and without infrastructure such as electricity and telecommunication, and due to being forbidden to wear or embed electronic tags onto wild rare animals, a new effective passive locating method has to be proposed, which must also face the problems, such as that a large number of nodes must be irregularly deployed in the huge area, and with large ambient noise, as well as the serious energy limitation. .The goal of the proposal is to investigate key techniques on passive localization of multi-moving targets in a three dimensional wild situation. The scale (or the whole number) of the moving targets should be estimated through machine learning and the model of RSSI signal degradation. A zone based layered model of passive localization is introduced to improve the efficiency of the localization, of which the inter-zone locating is achieved by RSSI matrix analysis, and the intra-zone localization is implemented using compressed sensing. A two-level scheduling model for node sleeping is to be used, and the wakeup strategy and the related coordination mechanisms among nodes within and/or between zones will be deeply studied. .The objective of the project is to explore the intrinsic relationships among node's topology of the WSN, the locating precision of multi-moving targets, and the robustness of the locating algorithms. A solution of passive localization of multi-moving targets in a large WSN with features of high adaptability, robustness, and energy-saving, will be finally achieved, which can be used in wildlife protection or even for other applications.

项目以无线传感网在野生动物行为监测中的应用为背景，针对珍稀野生动物不便或不许佩带标签的困难，及其活动环境人迹罕至、范围广、监测节点多、分布不规则、环境噪声大的特点，研究大型无线传感网中被动式多目标定位方法及其相关技术。试图采用分类学习与RSSI信号衰减模型相结合的方法进行目标规模检测，利用分层区域定位模型和压缩感知理论，在最小定位代价并保证精度条件下提高定位效率；利用自适应休眠调度方法优化大型网络生存期。特色在于利用压缩感知与基于RSSI感知矩阵特征结构分析方法和全局优化休眠调度策略，建立一个适应性强、鲁棒性好、资源节约的复杂三维环境中的多目标被动式定位方案。其科学实质是借助感知压缩、矩阵特征结构分析和机器学习理论，寻求复杂环境中无线信号与移动目标时空关系的内在规律，探索感知节点拓扑阵列与定位精度和算法鲁棒性的本质联系，问题的求解对复杂环境中大型无线传感网多目标定位研究及其应用有重要意义。

野生动物保护中，缺乏长期科学数据的积累，导致物种迁移流失、病害死亡、盗猎等无法及时发现。传感网目标定位技术能够有效对动物位置做出判定。野生动物活动区域大多为人迹罕至的环境，监测区域广、节点数量大、分布不规则、环境噪声大的特点，同时传感网本身能量有限，迫切需要研究新的适应大型野生动物行为监测需要的有效被动式移动目标规模判定方法、移动目标三维定位方法以及与之相适应的网络休眠调度策略。本项目设计了一种基于RSSI的轻量级目标规模判定方法，实现了对目标规模进行有效判定；项目引入分层分区域的思想，提高定位效率；设计了与定位策略相适应的两级休眠调度策略。项目的特色在于利用压缩传感思想和特征结构方法完成非规则、不均匀、复杂三维环境中的多目标被动式定位问题，解决能量和运算能力限制问题，建立了一个适应性强，鲁棒性好，节点资源节约的WSN优化定位方案，为大规模WSN在野生动物保护应用提供有价值参考。